telegabber/telegram/formatter/formatter.go

166 lines
4.5 KiB
Go

package formatter
import (
"sort"
log "github.com/sirupsen/logrus"
"github.com/zelenin/go-tdlib/client"
)
// Insertion is a piece of text in given position
type Insertion struct {
Offset int32
Runes []rune
}
// InsertionStack contains the sequence of insertions
// from the start or from the end
type InsertionStack []*Insertion
var boldRunes = []rune("**")
var italicRunes = []rune("_")
var codeRunes = []rune("\n```\n")
var urlRuneL = []rune("[")
// rebalance pumps all the values at given offset to current stack (growing
// from start) from given stack (growing from end); should be called
// before any insertions to the current stack at the given offset
func (s InsertionStack) rebalance(s2 InsertionStack, offset int32) (InsertionStack, InsertionStack) {
for len(s2) > 0 && s2[len(s2)-1].Offset <= offset {
s = append(s, s2[len(s2)-1])
s2 = s2[:len(s2)-1]
}
return s, s2
}
// NewIterator is a second order function that sequentially scans and returns
// stack elements; starts returning nil when elements are ended
func (s InsertionStack) NewIterator() func() *Insertion {
i := -1
return func() *Insertion {
i++
if i < len(s) {
return s[i]
}
return nil
}
}
// SortEntities arranges the entities in traversal-ready order
func SortEntities(entities []*client.TextEntity) []*client.TextEntity {
sortedEntities := make([]*client.TextEntity, len(entities))
copy(sortedEntities, entities)
sort.Slice(sortedEntities, func(i int, j int) bool {
entity1 := entities[i]
entity2 := entities[j]
if entity1.Offset < entity2.Offset {
return true
} else if entity1.Offset == entity2.Offset {
return entity1.Length > entity2.Length
}
return false
})
return sortedEntities
}
func markupBraces(entity *client.TextEntity, lbrace, rbrace []rune) (*Insertion, *Insertion) {
return &Insertion{
Offset: entity.Offset,
Runes: lbrace,
}, &Insertion{
Offset: entity.Offset + entity.Length,
Runes: rbrace,
}
}
// EntityToMarkdown generates the wrapping Markdown tags
func EntityToMarkdown(entity *client.TextEntity) (*Insertion, *Insertion) {
switch entity.Type.TextEntityTypeType() {
case client.TypeTextEntityTypeBold:
return markupBraces(entity, boldRunes, boldRunes)
case client.TypeTextEntityTypeItalic:
return markupBraces(entity, italicRunes, italicRunes)
case client.TypeTextEntityTypeCode, client.TypeTextEntityTypePre:
return markupBraces(entity, codeRunes, codeRunes)
case client.TypeTextEntityTypePreCode:
preCode, _ := entity.Type.(*client.TextEntityTypePreCode)
return markupBraces(entity, []rune("\n```"+preCode.Language+"\n"), codeRunes)
case client.TypeTextEntityTypeTextUrl:
textURL, _ := entity.Type.(*client.TextEntityTypeTextUrl)
return markupBraces(entity, urlRuneL, []rune("]("+textURL.Url+")"))
}
return nil, nil
}
// Format traverses an already sorted list of entities and wraps the text in Markdown
func Format(
sourceText string,
entities []*client.TextEntity,
entityToMarkup func(*client.TextEntity) (*Insertion, *Insertion),
) string {
if len(entities) == 0 {
return sourceText
}
startStack := make(InsertionStack, 0, len(sourceText))
endStack := make(InsertionStack, 0, len(sourceText))
// convert entities to a stack of brackets
var maxEndOffset int32
for _, entity := range entities {
log.Debugf("%#v", entity)
if entity.Length <= 0 {
continue
}
endOffset := entity.Offset + entity.Length
if endOffset > maxEndOffset {
maxEndOffset = endOffset
}
startStack, endStack = startStack.rebalance(endStack, entity.Offset)
startInsertion, endInsertion := entityToMarkup(entity)
if startInsertion != nil {
startStack = append(startStack, startInsertion)
}
if endInsertion != nil {
endStack = append(endStack, endInsertion)
}
}
// flush the closing brackets that still remain in endStack
startStack, endStack = startStack.rebalance(endStack, maxEndOffset)
// merge brackets into text
markupRunes := make([]rune, 0, len(sourceText))
nextInsertion := startStack.NewIterator()
insertion := nextInsertion()
var runeI int32
for _, cp := range sourceText {
for insertion != nil && insertion.Offset <= runeI {
markupRunes = append(markupRunes, insertion.Runes...)
insertion = nextInsertion()
}
markupRunes = append(markupRunes, cp)
// skip two UTF-16 code units (not points actually!) if needed
if cp > 0x0000ffff {
runeI += 2
} else {
runeI++
}
}
for insertion != nil {
markupRunes = append(markupRunes, insertion.Runes...)
insertion = nextInsertion()
}
return string(markupRunes)
}